fix(db): TypeORM 压缩配置对齐 SQL DDL，新增 TimescaleDB 初始化脚本
- kline.entity.ts: compress_segmentby 移除 interval（基表固定 1m 无需分段），schedule_interval 365d→30d 与 init-db SQL 一致 - data-source.ts: 生产环境关闭 synchronize，以 init-db SQL 脚本为建表唯一来源 - 新增 data/db/init-db/ 初始化 SQL 链： 01-timescaledb.sql — 启用 TimescaleDB 扩展 02-init-tables.sql — 核心业务表 + klines hypertable（7d chunk / 30d 压缩） 03-continuous-aggregates.sql — 分层连续聚合视图（5m→15m→30m→1h→4h→1d→1w）
This commit is contained in:
Rekey
@@ -15,7 +15,7 @@ export const AppDataSource = new DataSource({
|
||||
...Object.values(entities),
|
||||
],
|
||||
// 生产环境禁用 synchronize，使用 Migration
|
||||
synchronize: true,
|
||||
synchronize: false,
|
||||
migrations: [__dirname + "/migrations/*.{ts,js}"],
|
||||
// 连接池
|
||||
extra: {
|
||||
|
||||
@@ -7,7 +7,7 @@
|
||||
//
|
||||
// 关键 TimescaleDB 特性（由 @Hypertable 装饰器自动配置）：
|
||||
// - 自动按 time 列做时间分区（by_range）
|
||||
// - 列式压缩（compress），7 天后自动执行
|
||||
// - 列式压缩（compress），30 天后自动执行
|
||||
// - 通过 ContinuousAggregate 生成高周期 K 线视图
|
||||
//
|
||||
// 注意：@timescaledb/typeorm v0.0.1 为实验版本，
|
||||
@@ -40,9 +40,9 @@ import type { KlineInterval } from '../../types';
|
||||
compression: {
|
||||
compress: true,
|
||||
compress_orderby: "time DESC",
|
||||
compress_segmentby: "exchange, symbol, interval",
|
||||
compress_segmentby: "exchange, symbol",
|
||||
policy: {
|
||||
schedule_interval: "365 days", // 365 天后自动压缩
|
||||
schedule_interval: "30 days", // 30 天后自动压缩
|
||||
},
|
||||
},
|
||||
})
|
||||
|
||||
@@ -0,0 +1,30 @@
|
||||
-- ============================================================
|
||||
-- 01-timescaledb.sql — TimescaleDB 扩展初始化
|
||||
-- ============================================================
|
||||
-- Docker 容器首次启动时自动执行（/docker-entrypoint-initdb.d/）
|
||||
-- 确保 TimescaleDB 扩展在数据库级别启用。
|
||||
--
|
||||
-- init-db 完整执行链（按字母序自动执行）：
|
||||
-- 01-timescaledb.sql — 本文件：启用 TimescaleDB 扩展
|
||||
-- 02-init-tables.sql — 核心业务表（exchanges / trading_pairs / klines）
|
||||
-- 03-continuous-aggregates.sql — K 线分层连续聚合视图（5m → 1w）
|
||||
--
|
||||
-- 注意：
|
||||
-- - klines 基表由 02-init-tables.sql 创建为 TimescaleDB hypertable
|
||||
-- - 连续聚合视图由 03-continuous-aggregates.sql 创建
|
||||
-- - TypeORM 的 synchronize:true 与 SQL 脚本互为 fallback（开发/生产双路径）
|
||||
-- - 本脚本为 init-db 链的第一环，仅负责扩展启用
|
||||
-- ============================================================
|
||||
|
||||
-- 启用 TimescaleDB 扩展（必须最先执行）
|
||||
CREATE EXTENSION IF NOT EXISTS timescaledb CASCADE;
|
||||
|
||||
-- 验证扩展已启用
|
||||
DO $$
|
||||
BEGIN
|
||||
IF NOT EXISTS (
|
||||
SELECT 1 FROM pg_extension WHERE extname = 'timescaledb'
|
||||
) THEN
|
||||
RAISE EXCEPTION 'TimescaleDB extension is not installed';
|
||||
END IF;
|
||||
END $$;
|
||||
@@ -0,0 +1,344 @@
|
||||
-- ============================================================
|
||||
-- 02-init-tables.sql — 核心业务表建表语句
|
||||
-- ============================================================
|
||||
-- 根据 data/db/entities/ 中的实体定义生成对应 PostgreSQL DDL。
|
||||
--
|
||||
-- 表结构对应关系：
|
||||
-- exchanges ← exchange.entity.ts (Exchange extends CommonBaseEntity)
|
||||
-- trading_pairs ← trading-pair.entity.ts (TradingPair extends CommonBaseEntity)
|
||||
-- klines ← kline.entity.ts (Kline — TimescaleDB Hypertable)
|
||||
--
|
||||
-- 执行前提：
|
||||
-- 1. 01-timescaledb.sql 已执行（TimescaleDB 扩展已启用）
|
||||
-- 2. PostgreSQL 版本 >= 13（gen_random_uuid() 内建支持）
|
||||
--
|
||||
-- 幂等性：全程使用 IF NOT EXISTS / IF EXISTS，可重复执行。
|
||||
-- ============================================================
|
||||
|
||||
-- ============================================================
|
||||
-- 第一节：交易所配置表（exchanges）
|
||||
-- ============================================================
|
||||
-- 存储已接入的交易所元信息（Binance / OKX / Bybit 等）。
|
||||
-- 由 TypeORM Exchange 实体管理（关系数据域）。
|
||||
-- ============================================================
|
||||
|
||||
CREATE TABLE IF NOT EXISTS exchanges (
|
||||
-- UUID 主键（非自增整数，便于分布式场景）
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
|
||||
-- 交易所唯一标识（如 binance / okx / bybit）
|
||||
name VARCHAR(50) NOT NULL UNIQUE,
|
||||
|
||||
-- 交易所显示名称（如 Binance / OKX / Bybit）
|
||||
label VARCHAR(100) NOT NULL,
|
||||
|
||||
-- 是否启用该交易所的数据采集
|
||||
enabled BOOLEAN NOT NULL DEFAULT TRUE,
|
||||
|
||||
-- 交易所特定配置（JSON：费率、最小下单量、API 限频等）
|
||||
config JSONB,
|
||||
|
||||
-- 记录创建时间（自动填充）
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
|
||||
-- 最后更新时间（触发器自动刷新，见文末）
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
|
||||
-- 交易所名称查询索引
|
||||
CREATE INDEX IF NOT EXISTS idx_exchanges_name ON exchanges (name);
|
||||
-- 启用状态筛选索引
|
||||
CREATE INDEX IF NOT EXISTS idx_exchanges_enabled ON exchanges (enabled);
|
||||
|
||||
|
||||
-- ============================================================
|
||||
-- 第二节：交易对配置表（trading_pairs）
|
||||
-- ============================================================
|
||||
-- 存储各交易所的交易对元信息。数据模块启动时从该表读取
|
||||
-- active=true 的交易对列表，决定 WebSocket 订阅范围和 K 线合成范围。
|
||||
-- ============================================================
|
||||
|
||||
CREATE TABLE IF NOT EXISTS trading_pairs (
|
||||
-- UUID 主键
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
|
||||
-- 所属交易所（逻辑外键 → exchanges.id）
|
||||
exchange_id UUID NOT NULL
|
||||
REFERENCES exchanges(id) ON DELETE CASCADE,
|
||||
|
||||
-- 交易对符号（如 BTCUSDT / ETHUSDT）
|
||||
symbol VARCHAR(20) NOT NULL,
|
||||
|
||||
-- 基础币种（如 BTC）
|
||||
base_asset VARCHAR(10) NOT NULL,
|
||||
|
||||
-- 计价币种（如 USDT）
|
||||
quote_asset VARCHAR(10) NOT NULL,
|
||||
|
||||
-- 价格精度（小数位数）
|
||||
price_precision INTEGER NOT NULL DEFAULT 10,
|
||||
|
||||
-- 数量精度（小数位数）
|
||||
quantity_precision INTEGER NOT NULL DEFAULT 10,
|
||||
|
||||
-- 最小下单量
|
||||
min_qty NUMERIC(32, 8),
|
||||
|
||||
-- 下单步长（数量增量）
|
||||
step_size NUMERIC(32, 8),
|
||||
|
||||
-- 最小名义价值（USDT）
|
||||
min_notional NUMERIC(32, 8),
|
||||
|
||||
-- 是否激活数据订阅（false 时不采集该交易对行情）
|
||||
active BOOLEAN NOT NULL DEFAULT TRUE,
|
||||
|
||||
-- 是否启用 K 线合成（false 时仅采集原始行情，不合成）
|
||||
kline_synthesis_enabled BOOLEAN NOT NULL DEFAULT TRUE,
|
||||
|
||||
-- 默认 K 线周期
|
||||
kline_interval VARCHAR(100) NOT NULL DEFAULT '1m',
|
||||
|
||||
-- K 线合成周期列表（逗号分隔，如 "1m,5m,15m,1h,4h,1d"）
|
||||
kline_intervals VARCHAR(100) NOT NULL DEFAULT '1m,5m,15m,1h,4h,1d',
|
||||
|
||||
-- 历史 K 线最后补全时间（UTC）。默认 Unix epoch 起始，
|
||||
-- 新交易对从 epoch 起始时间开始全量补拉。
|
||||
last_backfill_time TIMESTAMPTZ NOT NULL DEFAULT to_timestamp(0),
|
||||
|
||||
-- 备注
|
||||
notes TEXT,
|
||||
|
||||
-- 审计时间戳
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
|
||||
-- 同一交易所下 symbol 唯一
|
||||
CONSTRAINT uq_trading_pairs_exchange_symbol UNIQUE (exchange_id, symbol)
|
||||
);
|
||||
|
||||
-- 按激活状态快速筛选
|
||||
CREATE INDEX IF NOT EXISTS idx_trading_pairs_active ON trading_pairs (active);
|
||||
-- 按交易所+交易对查询（最常用模式）
|
||||
CREATE INDEX IF NOT EXISTS idx_trading_pairs_exchange_symbol ON trading_pairs (exchange_id, symbol);
|
||||
|
||||
|
||||
-- ============================================================
|
||||
-- 第三节：1 分钟 K 线 Hypertable（klines）
|
||||
-- ============================================================
|
||||
-- TimescaleDB hypertable，存储交易所推送的 OHLCV 数据。
|
||||
-- 写入使用 UPSERT（ON CONFLICT DO UPDATE），已存在的 K 线
|
||||
-- 只更新 high/low/close/volume 增量。
|
||||
--
|
||||
-- TimescaleDB 配置：
|
||||
-- - chunk_time_interval: 7 days（周分区；1 day→7 days 减少 7× chunk 数）
|
||||
-- - 列式压缩：7 天后自动执行（压缩率 ~92%）
|
||||
-- - 压缩分段键：exchange, symbol（同交易对聚合压缩；interval 固定 1m 无需分段）
|
||||
-- - 压缩排序键：time DESC（查询通常按时间降序）
|
||||
--
|
||||
-- chunk 大小选择指南（16GB / i3-7300U / 1TB SSD）：
|
||||
-- interval chunk 数估算(1000万行) 单 chunk 行数 适用场景
|
||||
-- ─────────── ────────────────────── ───────────── ──────────────────
|
||||
-- 1 day 3200+（过碎 ❌） ~3,000 元数据开销 >> 数据
|
||||
-- 7 days ~450（推荐 ✅） ~22,000 查询剪枝 & 管理平衡
|
||||
-- 1 month ~100 ~100,000 历史归档为主、写入密集
|
||||
--
|
||||
-- 已有数据库在线修复（仅影响新 chunk，旧 chunk 需 migrate_chunk）：
|
||||
-- SELECT set_chunk_time_interval('klines', INTERVAL '7 days');
|
||||
-- ============================================================
|
||||
|
||||
CREATE TABLE IF NOT EXISTS klines (
|
||||
-- 交易所标识（binance / okx / bybit）
|
||||
exchange TEXT NOT NULL,
|
||||
|
||||
-- 交易对符号（如 BTCUSDT）
|
||||
symbol TEXT NOT NULL,
|
||||
|
||||
-- K 线周期（固定 "1m"，基表仅存 1 分钟）
|
||||
interval TEXT NOT NULL,
|
||||
|
||||
-- K 线开盘时间（UTC）— 时间分区键
|
||||
time TIMESTAMPTZ NOT NULL,
|
||||
|
||||
-- ============================================================
|
||||
-- OHLCV 价格数据
|
||||
--
|
||||
-- 类型选择：NUMERIC(20,8) vs DOUBLE PRECISION
|
||||
-- NUMERIC : 精确十进制，无浮点舍入；~10-13 字节/值，CPU 计算慢
|
||||
-- DOUBLE : IEEE 754 浮点；固定 8 字节/值，CPU 原生指令，快 3-5×
|
||||
--
|
||||
-- 对于 K 线价格数据，DOUBLE PRECISION 的 15 位有效数字完全够用
|
||||
-- （BTC @ $100K 量级精度到 $0.01 仅需 ~7 位有效数字）。
|
||||
-- 9 个价格列 × 1000 万行：NUMERIC → DOUBLE 可节省 ~200-400 MB 存储
|
||||
-- 并显著加速聚合/窗口函数。新部署强烈建议改为 DOUBLE PRECISION。
|
||||
-- ============================================================
|
||||
|
||||
-- 开盘价
|
||||
open NUMERIC(20, 8) NOT NULL,
|
||||
|
||||
-- 最高价
|
||||
high NUMERIC(20, 8) NOT NULL,
|
||||
|
||||
-- 最低价
|
||||
low NUMERIC(20, 8) NOT NULL,
|
||||
|
||||
-- 收盘价
|
||||
close NUMERIC(20, 8) NOT NULL,
|
||||
|
||||
-- 成交量（base 币种）
|
||||
volume NUMERIC(20, 8) NOT NULL,
|
||||
|
||||
-- ============================================================
|
||||
-- 扩展字段（连续聚合时使用 SUM 聚合）
|
||||
-- ============================================================
|
||||
|
||||
-- 成交额（quote 币种）
|
||||
quote_volume NUMERIC(20, 8),
|
||||
|
||||
-- 主动买入成交量（base 币种）
|
||||
taker_buy_base_vol NUMERIC(20, 8),
|
||||
|
||||
-- 主动买入成交额（quote 币种）
|
||||
taker_buy_quote_vol NUMERIC(20, 8),
|
||||
|
||||
-- 成交笔数
|
||||
trade_count INTEGER,
|
||||
|
||||
-- K 线是否已关闭（true = 该周期 K 线不再变化）
|
||||
is_closed BOOLEAN NOT NULL DEFAULT TRUE,
|
||||
|
||||
-- 审计时间戳
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
|
||||
-- 复合主键：同一交易所、同一交易对、同一周期、同一时间的 K 线唯一
|
||||
PRIMARY KEY (exchange, symbol, interval, time)
|
||||
);
|
||||
|
||||
-- ============================================================
|
||||
-- 将 klines 转换为 TimescaleDB Hypertable
|
||||
-- ============================================================
|
||||
|
||||
-- 创建 hypertable，按 time 列做周分区（chunk_time_interval = 7 days）
|
||||
-- 1 day → 7 days：chunk 数从 ~3200 降至 ~450，消除元数据瓶颈
|
||||
SELECT create_hypertable('klines', 'time',
|
||||
chunk_time_interval => INTERVAL '7 days',
|
||||
if_not_exists => TRUE
|
||||
);
|
||||
|
||||
-- ============================================================
|
||||
-- 配置列式压缩
|
||||
-- ============================================================
|
||||
|
||||
-- 启用列式压缩（先启用压缩，再设置分段/排序键）
|
||||
-- 注意：interval 在基表固定为 '1m'，从 segmentby 中移除以减少压缩分段数
|
||||
ALTER TABLE klines SET (
|
||||
timescaledb.compress,
|
||||
timescaledb.compress_segmentby = 'exchange,symbol',
|
||||
timescaledb.compress_orderby = 'time DESC'
|
||||
);
|
||||
|
||||
-- 添加压缩策略：30 天前的数据自动压缩
|
||||
-- chunk_time_interval=7d + compress_after=30d → 数据写入 ~37 天后被压缩
|
||||
-- 选择 30 天的理由：
|
||||
-- 量化交易中最常见的回测窗口是最近 30 天，保持未压缩可避免解压 CPU 开销。
|
||||
-- 30 天后的数据访问频率急剧下降，压缩带来的 IO 减少远超解压开销。
|
||||
-- 对于 4 币种：30 天未压缩 ≈ 36 MB（微不足道）
|
||||
-- 对于 100 币种：30 天未压缩 ≈ 907 MB（仍在 2GB shared_buffers 可接受范围）
|
||||
SELECT add_compression_policy('klines',
|
||||
compress_after => INTERVAL '30 days',
|
||||
if_not_exists => TRUE
|
||||
);
|
||||
|
||||
|
||||
-- ============================================================
|
||||
-- 第四节：updated_at 自动刷新触发器
|
||||
-- ============================================================
|
||||
-- TypeORM 的 @UpdateDateColumn 装饰器在应用层自动更新 updated_at。
|
||||
-- 但通过 SQL 直接操作表时（如补数据脚本），需通过触发器确保
|
||||
-- updated_at 在每次 UPDATE 时自动刷新到最新时间。
|
||||
--
|
||||
-- 注意：此触发器仅影响直接 SQL 操作，TypeORM 的 save()/update()
|
||||
-- 仍由其装饰器控制 updated_at 行为，两层互不冲突。
|
||||
-- ============================================================
|
||||
|
||||
CREATE OR REPLACE FUNCTION update_updated_at_column()
|
||||
RETURNS TRIGGER AS $$
|
||||
BEGIN
|
||||
NEW.updated_at = now();
|
||||
RETURN NEW;
|
||||
END;
|
||||
$$ LANGUAGE plpgsql;
|
||||
|
||||
-- 仅在触发器不存在时创建（幂等）
|
||||
DO $$
|
||||
BEGIN
|
||||
IF NOT EXISTS (
|
||||
SELECT 1 FROM pg_trigger
|
||||
WHERE tgname = 'trg_exchanges_updated_at'
|
||||
) THEN
|
||||
CREATE TRIGGER trg_exchanges_updated_at
|
||||
BEFORE UPDATE ON exchanges
|
||||
FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
|
||||
END IF;
|
||||
|
||||
IF NOT EXISTS (
|
||||
SELECT 1 FROM pg_trigger
|
||||
WHERE tgname = 'trg_trading_pairs_updated_at'
|
||||
) THEN
|
||||
CREATE TRIGGER trg_trading_pairs_updated_at
|
||||
BEFORE UPDATE ON trading_pairs
|
||||
FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
|
||||
END IF;
|
||||
|
||||
IF NOT EXISTS (
|
||||
SELECT 1 FROM pg_trigger
|
||||
WHERE tgname = 'trg_klines_updated_at'
|
||||
) THEN
|
||||
CREATE TRIGGER trg_klines_updated_at
|
||||
BEFORE UPDATE ON klines
|
||||
FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
|
||||
END IF;
|
||||
END $$;
|
||||
|
||||
|
||||
-- ============================================================
|
||||
-- 第五节：种子数据（可选）
|
||||
-- ============================================================
|
||||
-- 初始化默认交易所和常用交易对，方便开发环境快速启动。
|
||||
-- 生产环境请根据实际需求修改或删除此节。
|
||||
-- ============================================================
|
||||
|
||||
-- 默认交易所（幂等：ON CONFLICT DO NOTHING）
|
||||
INSERT INTO exchanges (name, label, enabled, config) VALUES
|
||||
('binance', 'Binance', TRUE,
|
||||
'{"rateLimit": 1200, "minOrderSize": 0.001, "feeTaker": 0.001, "feeMaker": 0.001}'::jsonb),
|
||||
('okx', 'OKX', TRUE,
|
||||
'{"rateLimit": 400, "minOrderSize": 0.001, "feeTaker": 0.001, "feeMaker": 0.0008}'::jsonb),
|
||||
('bybit', 'Bybit', FALSE,
|
||||
'{"rateLimit": 600, "minOrderSize": 0.001, "feeTaker": 0.001, "feeMaker": 0.001}'::jsonb)
|
||||
ON CONFLICT (name) DO NOTHING;
|
||||
|
||||
-- 默认交易对（仅 Binance 主流 USDT 永续合约，幂等）
|
||||
INSERT INTO trading_pairs (exchange_id, symbol, base_asset, quote_asset,
|
||||
price_precision, quantity_precision, kline_interval, kline_intervals, active)
|
||||
SELECT
|
||||
e.id,
|
||||
sym.symbol,
|
||||
sym.base,
|
||||
sym.quote,
|
||||
2, -- price_precision（USDT 计价通常 2 位小数）
|
||||
5, -- quantity_precision（数量通常 5 位小数）
|
||||
'1m',
|
||||
'1m,5m,15m,30m,1h,4h,1d,1w',
|
||||
TRUE
|
||||
FROM exchanges e
|
||||
CROSS JOIN (
|
||||
VALUES
|
||||
('BTCUSDT', 'BTC', 'USDT'),
|
||||
('ETHUSDT', 'ETH', 'USDT'),
|
||||
('BNBUSDT', 'BNB', 'USDT'),
|
||||
('SOLUSDT', 'SOL', 'USDT')
|
||||
) AS sym(symbol, base, quote)
|
||||
WHERE e.name = 'binance'
|
||||
ON CONFLICT (exchange_id, symbol) DO NOTHING;
|
||||
@@ -0,0 +1,277 @@
|
||||
-- ============================================================
|
||||
-- 03-continuous-aggregates.sql — K 线分层连续聚合视图
|
||||
-- ============================================================
|
||||
-- 从 klines（1m）基表创建分层连续聚合物化视图链：
|
||||
-- 1m → 5m → 15m → 30m → 1h → 4h → 1d → 1w
|
||||
--
|
||||
-- 执行前提：
|
||||
-- 1. klines hypertable 已创建（由 02-init-tables.sql 创建）
|
||||
-- 2. klines 表中已有数据（至少一条，否则视图创建成功但无数据）
|
||||
--
|
||||
-- 执行方式：
|
||||
-- psql -U trader -d trade -f 03-continuous-aggregates.sql
|
||||
--
|
||||
-- 幂等性：使用 IF NOT EXISTS，可重复执行
|
||||
--
|
||||
-- ============================================================
|
||||
-- 聚合刷新模式选择（二选一）
|
||||
-- ============================================================
|
||||
-- 本脚本默认注释掉所有 add_continuous_aggregate_policy 调用。
|
||||
-- 你需要根据部署场景选择一种刷新模式：
|
||||
--
|
||||
-- 【模式 A：定时调度刷新】（传统方式，适合简单场景）
|
||||
-- 取消注释各节的 add_continuous_aggregate_policy 调用即可。
|
||||
-- TimescaleDB Job Scheduler 按 schedule_interval 自动刷新。
|
||||
-- 缺点：回填期间可能与 INSERT 竞争资源；聚合有调度延迟。
|
||||
--
|
||||
-- 【模式 B：应用层触发式刷新】（推荐，精细控制）
|
||||
-- 保持 policy 注释状态。在应用层写入每条 1m K 线后，
|
||||
-- 检测时间桶是否关闭，若关闭则调用 refresh_continuous_aggregate。
|
||||
-- 优点：回填零干扰；聚合零延迟（桶关闭立即刷新）；无调度开销。
|
||||
-- 应用层代码模板见 04-backfill-workflow.sql 末尾。
|
||||
--
|
||||
-- 回填工作流（两种模式通用）：
|
||||
-- 1. 批量 INSERT 历史 K 线（policy 已注释，不冲突）
|
||||
-- 2. 手动全量刷新所有视图（见文件末尾注释块）
|
||||
-- 3. 接入实时数据（模式 A 启用 policy / 模式 B 应用层触发）
|
||||
-- ============================================================
|
||||
|
||||
-- ============================================================
|
||||
-- 5m K 线（从 1m 基表聚合）
|
||||
-- ============================================================
|
||||
CREATE MATERIALIZED VIEW IF NOT EXISTS klines_5m
|
||||
WITH (timescaledb.continuous) AS
|
||||
SELECT
|
||||
time_bucket('5 minutes', time) AS time,
|
||||
exchange,
|
||||
symbol,
|
||||
'5m'::text AS interval,
|
||||
FIRST(open, time) AS open,
|
||||
MAX(high) AS high,
|
||||
MIN(low) AS low,
|
||||
LAST(close, time) AS close,
|
||||
SUM(volume) AS volume,
|
||||
SUM(quote_volume) AS quote_volume,
|
||||
SUM(taker_buy_base_vol) AS taker_buy_base_vol,
|
||||
SUM(taker_buy_quote_vol) AS taker_buy_quote_vol,
|
||||
SUM(trade_count)::integer AS trade_count
|
||||
FROM klines
|
||||
GROUP BY time_bucket('5 minutes', klines.time), exchange, symbol
|
||||
WITH NO DATA;
|
||||
|
||||
-- 【模式 A 用户】取消下面注释以启用定时调度刷新
|
||||
-- SELECT add_continuous_aggregate_policy('klines_5m',
|
||||
-- start_offset => INTERVAL '1 day',
|
||||
-- end_offset => INTERVAL '5 minutes',
|
||||
-- schedule_interval => INTERVAL '5 minutes',
|
||||
-- if_not_exists => TRUE
|
||||
-- );
|
||||
|
||||
-- ============================================================
|
||||
-- 15m K 线（从 5m 聚合，分层链）
|
||||
-- ============================================================
|
||||
CREATE MATERIALIZED VIEW IF NOT EXISTS klines_15m
|
||||
WITH (timescaledb.continuous) AS
|
||||
SELECT
|
||||
time_bucket('15 minutes', time) AS time,
|
||||
exchange,
|
||||
symbol,
|
||||
'15m'::text AS interval,
|
||||
FIRST(open, time) AS open,
|
||||
MAX(high) AS high,
|
||||
MIN(low) AS low,
|
||||
LAST(close, time) AS close,
|
||||
SUM(volume) AS volume,
|
||||
SUM(quote_volume) AS quote_volume,
|
||||
SUM(taker_buy_base_vol) AS taker_buy_base_vol,
|
||||
SUM(taker_buy_quote_vol) AS taker_buy_quote_vol,
|
||||
SUM(trade_count)::integer AS trade_count
|
||||
FROM klines_5m
|
||||
GROUP BY time_bucket('15 minutes', klines_5m.time), exchange, symbol
|
||||
WITH NO DATA;
|
||||
|
||||
-- 【模式 A 用户】取消下面注释以启用定时调度刷新
|
||||
-- SELECT add_continuous_aggregate_policy('klines_15m',
|
||||
-- start_offset => INTERVAL '2 days',
|
||||
-- end_offset => INTERVAL '15 minutes',
|
||||
-- schedule_interval => INTERVAL '15 minutes',
|
||||
-- if_not_exists => TRUE
|
||||
-- );
|
||||
|
||||
-- ============================================================
|
||||
-- 30m K 线（从 15m 聚合，分层链）
|
||||
-- ============================================================
|
||||
CREATE MATERIALIZED VIEW IF NOT EXISTS klines_30m
|
||||
WITH (timescaledb.continuous) AS
|
||||
SELECT
|
||||
time_bucket('30 minutes', time) AS time,
|
||||
exchange,
|
||||
symbol,
|
||||
'30m'::text AS interval,
|
||||
FIRST(open, time) AS open,
|
||||
MAX(high) AS high,
|
||||
MIN(low) AS low,
|
||||
LAST(close, time) AS close,
|
||||
SUM(volume) AS volume,
|
||||
SUM(quote_volume) AS quote_volume,
|
||||
SUM(taker_buy_base_vol) AS taker_buy_base_vol,
|
||||
SUM(taker_buy_quote_vol) AS taker_buy_quote_vol,
|
||||
SUM(trade_count)::integer AS trade_count
|
||||
FROM klines_15m
|
||||
GROUP BY time_bucket('30 minutes', klines_15m.time), exchange, symbol
|
||||
WITH NO DATA;
|
||||
|
||||
-- 【模式 A 用户】取消下面注释以启用定时调度刷新
|
||||
-- SELECT add_continuous_aggregate_policy('klines_30m',
|
||||
-- start_offset => INTERVAL '3 days',
|
||||
-- end_offset => INTERVAL '30 minutes',
|
||||
-- schedule_interval => INTERVAL '30 minutes',
|
||||
-- if_not_exists => TRUE
|
||||
-- );
|
||||
|
||||
-- ============================================================
|
||||
-- 1h K 线（从 30m 聚合，分层链）
|
||||
-- ============================================================
|
||||
CREATE MATERIALIZED VIEW IF NOT EXISTS klines_1h
|
||||
WITH (timescaledb.continuous) AS
|
||||
SELECT
|
||||
time_bucket('1 hour', time) AS time,
|
||||
exchange,
|
||||
symbol,
|
||||
'1h'::text AS interval,
|
||||
FIRST(open, time) AS open,
|
||||
MAX(high) AS high,
|
||||
MIN(low) AS low,
|
||||
LAST(close, time) AS close,
|
||||
SUM(volume) AS volume,
|
||||
SUM(quote_volume) AS quote_volume,
|
||||
SUM(taker_buy_base_vol) AS taker_buy_base_vol,
|
||||
SUM(taker_buy_quote_vol) AS taker_buy_quote_vol,
|
||||
SUM(trade_count)::integer AS trade_count
|
||||
FROM klines_30m
|
||||
GROUP BY time_bucket('1 hour', klines_30m.time), exchange, symbol
|
||||
WITH NO DATA;
|
||||
|
||||
-- 【模式 A 用户】取消下面注释以启用定时调度刷新
|
||||
-- SELECT add_continuous_aggregate_policy('klines_1h',
|
||||
-- start_offset => INTERVAL '7 days',
|
||||
-- end_offset => INTERVAL '1 hour',
|
||||
-- schedule_interval => INTERVAL '1 hour',
|
||||
-- if_not_exists => TRUE
|
||||
-- );
|
||||
|
||||
-- ============================================================
|
||||
-- 4h K 线（从 1h 聚合，分层链）
|
||||
-- ============================================================
|
||||
CREATE MATERIALIZED VIEW IF NOT EXISTS klines_4h
|
||||
WITH (timescaledb.continuous) AS
|
||||
SELECT
|
||||
time_bucket('4 hours', time) AS time,
|
||||
exchange,
|
||||
symbol,
|
||||
'4h'::text AS interval,
|
||||
FIRST(open, time) AS open,
|
||||
MAX(high) AS high,
|
||||
MIN(low) AS low,
|
||||
LAST(close, time) AS close,
|
||||
SUM(volume) AS volume,
|
||||
SUM(quote_volume) AS quote_volume,
|
||||
SUM(taker_buy_base_vol) AS taker_buy_base_vol,
|
||||
SUM(taker_buy_quote_vol) AS taker_buy_quote_vol,
|
||||
SUM(trade_count)::integer AS trade_count
|
||||
FROM klines_1h
|
||||
GROUP BY time_bucket('4 hours', klines_1h.time), exchange, symbol
|
||||
WITH NO DATA;
|
||||
|
||||
-- 【模式 A 用户】取消下面注释以启用定时调度刷新
|
||||
-- SELECT add_continuous_aggregate_policy('klines_4h',
|
||||
-- start_offset => INTERVAL '14 days',
|
||||
-- end_offset => INTERVAL '4 hours',
|
||||
-- schedule_interval => INTERVAL '4 hours',
|
||||
-- if_not_exists => TRUE
|
||||
-- );
|
||||
|
||||
-- ============================================================
|
||||
-- 1d K 线（从 4h 聚合，分层链）
|
||||
-- ============================================================
|
||||
CREATE MATERIALIZED VIEW IF NOT EXISTS klines_1d
|
||||
WITH (timescaledb.continuous) AS
|
||||
SELECT
|
||||
time_bucket('1 day', time) AS time,
|
||||
exchange,
|
||||
symbol,
|
||||
'1d'::text AS interval,
|
||||
FIRST(open, time) AS open,
|
||||
MAX(high) AS high,
|
||||
MIN(low) AS low,
|
||||
LAST(close, time) AS close,
|
||||
SUM(volume) AS volume,
|
||||
SUM(quote_volume) AS quote_volume,
|
||||
SUM(taker_buy_base_vol) AS taker_buy_base_vol,
|
||||
SUM(taker_buy_quote_vol) AS taker_buy_quote_vol,
|
||||
SUM(trade_count)::integer AS trade_count
|
||||
FROM klines_4h
|
||||
GROUP BY time_bucket('1 day', klines_4h.time), exchange, symbol
|
||||
WITH NO DATA;
|
||||
|
||||
-- 【模式 A 用户】取消下面注释以启用定时调度刷新
|
||||
-- SELECT add_continuous_aggregate_policy('klines_1d',
|
||||
-- start_offset => INTERVAL '30 days',
|
||||
-- end_offset => INTERVAL '1 day',
|
||||
-- schedule_interval => INTERVAL '1 day',
|
||||
-- if_not_exists => TRUE
|
||||
-- );
|
||||
|
||||
-- ============================================================
|
||||
-- 1w K 线（从 1d 聚合，分层链）
|
||||
-- ============================================================
|
||||
CREATE MATERIALIZED VIEW IF NOT EXISTS klines_1w
|
||||
WITH (timescaledb.continuous) AS
|
||||
SELECT
|
||||
time_bucket('1 week', time) AS time,
|
||||
exchange,
|
||||
symbol,
|
||||
'1w'::text AS interval,
|
||||
FIRST(open, time) AS open,
|
||||
MAX(high) AS high,
|
||||
MIN(low) AS low,
|
||||
LAST(close, time) AS close,
|
||||
SUM(volume) AS volume,
|
||||
SUM(quote_volume) AS quote_volume,
|
||||
SUM(taker_buy_base_vol) AS taker_buy_base_vol,
|
||||
SUM(taker_buy_quote_vol) AS taker_buy_quote_vol,
|
||||
SUM(trade_count)::integer AS trade_count
|
||||
FROM klines_1d
|
||||
GROUP BY time_bucket('1 week', klines_1d.time), exchange, symbol
|
||||
WITH NO DATA;
|
||||
|
||||
-- 【模式 A 用户】取消下面注释以启用定时调度刷新
|
||||
-- SELECT add_continuous_aggregate_policy('klines_1w',
|
||||
-- start_offset => INTERVAL '90 days',
|
||||
-- end_offset => INTERVAL '1 day',
|
||||
-- schedule_interval => INTERVAL '1 day',
|
||||
-- if_not_exists => TRUE
|
||||
-- );
|
||||
|
||||
-- ============================================================
|
||||
-- 推荐索引：加速按 symbol + time 的查询
|
||||
-- ============================================================
|
||||
CREATE INDEX IF NOT EXISTS idx_klines_5m_symbol_time ON klines_5m (exchange, symbol, time DESC);
|
||||
CREATE INDEX IF NOT EXISTS idx_klines_15m_symbol_time ON klines_15m (exchange, symbol, time DESC);
|
||||
CREATE INDEX IF NOT EXISTS idx_klines_30m_symbol_time ON klines_30m (exchange, symbol, time DESC);
|
||||
CREATE INDEX IF NOT EXISTS idx_klines_1h_symbol_time ON klines_1h (exchange, symbol, time DESC);
|
||||
CREATE INDEX IF NOT EXISTS idx_klines_4h_symbol_time ON klines_4h (exchange, symbol, time DESC);
|
||||
CREATE INDEX IF NOT EXISTS idx_klines_1d_symbol_time ON klines_1d (exchange, symbol, time DESC);
|
||||
CREATE INDEX IF NOT EXISTS idx_klines_1w_symbol_time ON klines_1w (exchange, symbol, time DESC);
|
||||
|
||||
-- ============================================================
|
||||
-- 首次创建后手动刷新所有视图（填充历史数据）
|
||||
-- 取消注释以下行执行：
|
||||
-- ============================================================
|
||||
-- CALL refresh_continuous_aggregate('klines_5m', NULL, NULL);
|
||||
-- CALL refresh_continuous_aggregate('klines_15m', NULL, NULL);
|
||||
-- CALL refresh_continuous_aggregate('klines_30m', NULL, NULL);
|
||||
-- CALL refresh_continuous_aggregate('klines_1h', NULL, NULL);
|
||||
-- CALL refresh_continuous_aggregate('klines_4h', NULL, NULL);
|
||||
-- CALL refresh_continuous_aggregate('klines_1d', NULL, NULL);
|
||||
-- CALL refresh_continuous_aggregate('klines_1w', NULL, NULL);
|
||||
Reference in New Issue
Block a user